SPV: Emit op select

Also, ensures it has a type, no disallowed side effects,
or performance trade offs.

HLSL: Wrap entry-point; needed to write 'in' args, and support 'inout' args

This enables the IO type mapping to work transparently for typedefs.

Previously, this was done recursively, per object, and the nonIO version
was cached. This reverses both those approaches.

This removes pervertex output blocks, in favor of using only
loose variables.  The pervertex blocks are not required and were
only partly implemented, and were adding some complication.

This change goes with wrap-entry-point.

Structs are split to remove builtin members to create valid SPIR-V.  In this
process, an outer structure array dimension may be propegated onto the
now-removed builtin variables.  For example, a mystruct[3].position ->
position[3].  The copy between the split and unsplit forms would handle
this in some cases, but not if the array dimension was at different levels
of aggregate.

It now does this, but may not handle arbitrary composite types.  Unclear if
that has any semantic meaning for builtins though.

This also removes an no longer needed makeTemporary() and rationalizes
makeTypeNonIo()'s interface.

This introduces parallel types for IO-type containing aggregates used as
non-entry point function parameters or return types, or declared as variables.
Further uses of the same original type will share the same sanitized deep
structure.

This is intended to be used with the wrap-entry-point branch.

Previously, a type graph would turn into a type tree. That is,
a deep node that is shared would have multiple copies made.

This is important when creating IO and non-IO versions of deep types.

This needs some render testing, but is destined to be part of master.

This also leads to a variety of other simplifications.
 - IO are global symbols, so only need one list of linkage nodes (deferred)
 - no longer need parse-context-wide 'inEntryPoint' state, entry-point is localized
 - several parts of splitting/flattening are now localized

SPV: Emit names of specialization constants, fixes #710

Install headers

This enables the vulkan loader to be built against an installed glslang.

CMake: add target_link_libraries internally

SpvBuilder: add const specifier to vector reference arguments

Use correct OriginUpperLeft when linking more than one shader.

Semantics: Texture clamp functions are available for fragment shader only

Fix a CMakeLists issue

Fix build if NV_EXTENSIONS is not set

support SPV_NV_viewport_array2 and SPV_NV_stereo_view_rendering

Makes it easier to include glslang in a larger CMake project---instead
of having to call `target_link_libraries(glslang OSDependent OGLCompiler
HLSL)`, for example, you only need to call
`target_link_libraries(glslang)` and it will pull in the helpers it
needs.

This is also better in terms of cleaning up the "public interface",
of sorts, for building glslang: end-users probably shouldn't need to
know or be explicitly dependent on internal targets.